difference between histcounts and imhist matlab-CodePudding

histcounts and imhist are not returning the same values for counts and bin locations.

x = [0.5000, 0.6429, 0.7143, 0.6429, 0.7857, 0.2857, 0.8571, 0.6429,0, 0.7857, 0.9286, 1.0000, 0.1429, 0.8571, 0.2857, 0.8571, 0.5714, 0.0714];

[c1, l1] = histcounts(x, 6)

c1 =
     3     2     1     4     3     5

l1 =
         0    0.1700    0.3400    0.5100    0.6800    0.8500    1.0200


[c2, l2] = imhist(x, 6)

c2 =
     2     3     0     5     6     2

l2 = 
         0    0.2000    0.4000    0.6000    0.8000    1.0000

What could be the reason for that?

CodePudding user response：

MATLAB

close all;clear all;clc

nbins=[6 16 26 46]

x = [0.5000, 0.6429, 0.7143, 0.6429, 0.7857, 0.2857, 0.8571, ...
    0.6429,0, 0.7857, 0.9286, 1.0000, 0.1429, 0.8571, 0.2857, 0.8571, 0.5714, 0.0714];

one can take it from one side

for k=1:1:numel(nbins)
    figure(k);
    ax=gca;hold on;grid on
    [C1, L1] = histcounts(x,nbins(k));
    stem(L1(1:end-1),C1);hold on
    [C2, L2] = imhist(x,nbins(k));
    stem(ax,L2,C2)

end

or from the other, stem graphs not shown, quite similar to the above ones.

for k=1:1:numel(nbins)
    figure(k);
    ax=gca;hold on;grid on
    [C1, L1] = histcounts(x,nbins(k));
    stem(L1(2:end),C1);hold on
    [C2, L2] = imhist(x,nbins(k));
    stem(ax,L2,C2)

end

The point : imhist is a command for images and it applies an offset to all histogram bin locations depending upon the type of image fed in.

imhist doesn't have a cut-off for tiny images, so the sequence x is assumed as image, which it is not.

Read imhist